def test_mbart50():
mt = dlt.TranslationModel("mbart50")
msg_en = "Hello everyone, how are you?"
fr_1 = mt.translate(msg_en, source="English", target="French")
ch = mt.translate(msg_en, source="English", target="Chinese")
fr_2 = mt.translate([msg_en, msg_en + msg_en],
source="English",
target="French")
assert fr_1 == fr_2[0]
assert ch != fr_1
def test_translate():
mt = dlt.TranslationModel()
msg_en = "Hello everyone, how are you?"
assert (mt.translate(msg_en, source="English",
target="Spanish") == "Hola a todos, ¿cómo estás?")
fr_1 = mt.translate(msg_en, source="English", target="French")
ch = mt.translate(msg_en, source="English", target="Chinese")
fr_2 = mt.translate([msg_en, msg_en + msg_en],
source="English",
target="French")
assert fr_1 == fr_2[0]
assert ch != fr_1
from auth import api_key
from flask import Flask
from flask import Blueprint
from flask import request
# import test as test
# import translate as translator
import ocror as ocror
import detectlang as detect
import os
# os.environ['TF_CPP_MIN_LOG_LEVEL'] = '3' # or any {'0', '1', '2'}
# import tensorflow as tf
import dl_translate as dlt
app = Flask(__name__)
app.secret_key = '_5#y2LF4Q8z]/'
mt = dlt.TranslationModel()
lang_code_map = mt.get_lang_code_map()
code_lang_map = {v: k for k, v in lang_code_map.items()}
from functools import wraps
from flask import g, request, jsonify, redirect, url_for
def api_key(f):
@wraps(f)
def decorator(*args, **kwargs):
token = None
if 'x-api-key' in request.headers:
token = request.headers['x-api-key']
# Prints input, output and evaluation line by line
def print_results(model, src_lang, tgt_lang, input_text):
output = spin(model=model,
src_lang=src_lang,
tgt_lang=tgt_lang,
input_text=input_text)
print(
"-------------------------------------------------------------------------------------------------------"
)
print("Input: " + input_text)
print("Output: " + output)
print("Spun: " + str(evaluate_output(input_text, output)))
models = [dlt.TranslationModel("mbart50"), dlt.TranslationModel("m2m100")]
src_lang = "German"
tgt_lang = "English"
text_list = []
rnd_list = []
text_dict = {'text': [], 'random': []}
corpusfile = open('res\en\eng_news_2020_10K\eng_news_2020_10K-sentences.txt',
'r',
encoding='utf-8')
for line in corpusfile.readlines():
text_list.append(line.split(
"\t")[1].rstrip()) # remove leading number, tab and trailing line feed
# print(line.split("\t")[1])
for text in text_list:
length = len(lang_list)
for i in range(length):
if i < length-1:
print(i)
print(lang_list[i])
print(lang_list)
intermediate = model.translate(intermediate,
source=lang_list[i],
target=lang_list[i+1],
batch_size=32,
generation_options=options)
return intermediate
#models = [dlt.TranslationModel("mbart50"), dlt.TranslationModel("m2m100")]
models = [dlt.TranslationModel("mbart50")]
language_combinations_de = [
['German', 'English', 'German'],
['German', 'Portuguese', 'German'],
['German', 'English', 'Portuguese', 'German']
]
language_combinations_en = [
['English', 'German', 'English'],
['English', 'Portuguese', 'English'],
['English', 'German', 'Portuguese', 'English']
]
opt_list = [dict(num_beams=1),
dict(num_beams=10),
dict(num_beams=1, do_sample=True, top_k=0, temperature=0.7),
dict(num_beams=1, do_sample=True, top_k=0, temperature=0.3)]
opt1 = dict(num_beams=1)